Taught in 2021–2022 by Benoît Crabbé and Philippe de Groote.
First semester course, on Tuesdays 12:45–15:45 in room 1004. Classes delivered in English if at least one student asks for it; in French otherwise.
We sketch here the planned contents for 2021–2022.
These contents are structured around three important subdomains of
linguistics, (morphology, syntax, and semantics),
presenting on each occasion some of the related models and the
corresponding algorithmic issues. The exact dates and content might
change.
-
General Introduction Language has structure. Language and inference. The importance of ambiguity. Language and the world.
Linguistics basics for computational linguistics. Statistical properties of words, constituent and dependency analyses, computing semantic denotations and computing semantic similarities.
Machine learning basics for computational linguistics. Coding discrete symbols as vectors (word embeddings), optimisation reminder.
-
Modelling sequences Presentation of typical problems involving sequence modelling.
Generative models language models, hidden markov models, PCFG
Discriminative models conditional random fields
Algorithms Viterbi and approximative methods
Deep learning based methods
October 5th, 2021
Modelling syntax (
slides)
Phrase structure grammar
Tree adjoining Grammar
Dependency syntax
Categorial grammar
October 19th, 2021
Parsing algorithms for natural language (
slides)
CKY and Earley Introduction to weighted CKY and Earley
Shift Reduce and Eisner for Dependency syntax
CKY for tree adjoining grammar
October 26th, 2021
November 2nd, 2021
November 9th, 2021
November 16th, 2021 Discourse Analysis discourse representation theory, anaphora resolution, type-theoretic dynamic logic
November 23rd, 2021